REACTIONS OF TRIPHENYLPHOSPHINE 1595
Vol. 34, No. 6, June 1069 tilled under vacuum. The temperature of the oil bath ww gradually raised t o 160'. A forerun, 20 g, bp 39" (0.15 mm), was followed by the main fraction which afforded 132.7 g (96.3%) of X : bp 42-44" (0.15 mm); n20~1.5324. The ir spectrum was superimposable with that of X, obtained above. C. With Triphenylphosphine in Ether.-A solution of IX, 34.8 g (0.1 mol), in 50 ml of ether was added dropwise with stirring to a solution of triphenylphosphine, 52.4 g (0.2 mol), in 1000 ml of ether at 25-35'. After the addition was complete, no starting material could be detected in solution by means of glpc. Water, 500 ml, was added dropwise with stirring. Triphenylphosphine oxide precipitated out shortly afterward and was removed by suction-filtration. The ether layer was dried (CaC12) and evaporated to dryness. The crude product was purified by distillation to give 39.8 g (82.1%) of 1,1,2,4,4-pentachloro-1-buten-3-one (XII): bp 89" (9 mm) ; n20~ 1.5440: dZ041.6628 [lit.24 bp 99--100" (13 mm); n% 1.44421; ir (film) 5.9 ( C 4 ) and 6.5 p (C=O) ; uv max 212 mp ( E 7953) and 271 (3071) ; nmr 6 6.76. Anal. Calcd for C4HC160: C, 19.8; H, 0.4; C1,73.2. Found: C, 20.0; H, 0.7; C1, 73.2. l-Phenyl-4,5-dichloro-3-dichloromethylpyazole(XIV) .-To a solution of 24.25 g (0.1 mol) of XI1 in 300 ml of ether was added dropwise a solution of '21.6 g (0.2 mol) of phenylhydrazine in 50 ml of ether. After 30 min 200 ml of water was added. The organic layer was dried (NapS04)and evaporated to dryness. The crude pyrazole was purified by distillation under vacuum to give 12.43 g ( 4 2 5 ) of pure XIV, which solidified: bp 142" (0.4 m m ) ; mp 53-55.5" [lit.*l mp 57.558"; bp 155" (0.8 mm)]; uv max 247 mp ( e 11,8166); nmr S 7.00 (aliphatic H) and 7.45 (aromatic H ) . Anal. Calcd for CloH6ClaN2: C, 40.5; H, 2.0; C1, 47.9; N, 9.5. Found: C, 40.4; H, 2.2; C1, 47.5; N, 9.7. (24) A. Roedig and H. J. 13ecker, Ber., 89, 906 (1956).
Reaction of Octachlorobutanone (IX) and 1,1,2,4,4,4-Hexachloro- 1-buten-3-one (X)with Trialkyl Phosphites.-The results are summarized in Table I. The general procedures are illustrated by the reactions of IX and X with trimethyl phosphite. A.-To a stirred solution of IX, 34.8 g (0.1 mol), in 50 ml of ether WM added dropwise trimethyl phosphite, 24.8 g (0.2 mol). The reaction was exothermic and the ether started to boil. The reaction mixture was heated to reflux for 10 min. Ether was removed under vacuum and the residual oil was distilled to give 13.0 g (90%) of dimethyl phosphorochloridate (IVa), bp 60-62" (8 mm), identical (glpc) with the product obtained by dechlorination of I with trimethyl phosphite (see above). The vinyl phosphate XVI (R = CHI) distilled at 88-89" (0.01 mm) : 26.6 g (76%) ; % * O D 1.5173; d 3 1.5869. The ester is a colorless liquid and turns light yellow on storage for several weeks when exposed to sunlight: uv max 208 mp ( e 20,088) and max 247 (7375); ir (film) 6.35 (C=C) and 7.7 p (P+O). B.-To a stirred solution of X, 10.65 g (0.0385 mol) in 20 ml of ether, was added trimethyl phosphite, 6.2 g (0.05 mol), dissolved in 20 ml of ether. The reaction was exothermic. Ether was removed and the residual oil was purified by distillation to give 12.9 g (95.57c) of XVI (R = CHI) : nZoD1.5163; d2o4 1.5899; uv and ir spectrum were identical with the product obtained in A.
Registry No.-11, 13340-09-5; IVa, 813-77-4; V, 18791-16-7; VIIa, 18766-86-4; VIIb, 18791-17-8; X, 13340-11-9; XII, 13340-10-8; XIV, 18767-09-4; XVI (R = Me), 18767-10-7; XVI (R = CzHs), 18767-11-8; XVI [R = CH(CH,)z], 18767-12-9; XVI (R = CHZCH=CH,), 18767-13-0; XVI (R = n-C4H9), 18767-141; XVI (R = ?&H11), 18791-18-9. 18767-15-2; XVI (R = ?%-C6H13),
The Reactions of Triphenylphosphine with a-Halobenzyl Phenyl Ketones and with a-Mesyloxybenzyl Phenyl Ketone' IRVING J. BOROWITZ,~ PAULE. RUSEK,AND REIN VIRKHAUS Department of Chemistry, Lehigh University, Bethlehem, Pennsylvania, and the Belfer GTadwlte School of Science, Yeshiva University, New York, New York loOS3 Received October 21, 1968 The reactions of a-bromobenzyl phenyl ketone (11) and a-chlorobenzyl phenyl ketone (14) with triphenylphosphine are described. Both enol phosphonium and ketophosphonium halides are formed in ratios dependent upon reaction conditions. The enol phosphonium halides are solvolyzed to deoxybenzoin (16) and converted into diphenylacetylene (17). Debromination of 11 to 16 occurs with triphenylphosphine in the presence of methanol. a-Mesyloxybenzyl phenyl ketone (19) reacts with triphenylphosphine to give only the a-ketophosphonium mesylate via displacement of mesylate ion. Probable mechanisms for the observed reactions and the relationships of these reactions to the reactions of other a-halo ketones with phosphines are discussed.
Recent work has shown that the reactions of triphenylphosphine with a-bromoacetophenone (1) and with a-bromopropiophenone (2) give the corresponding a-ketotriphenylphosphonium bromides in aprotic solv e n t ~ . Our ~ ~ kinetic ~ studies indicate that both 1 and 2 probably react with triphenylphosphine via displacement of bromide ion under aprotic condition^.^,^ We and others have previously postulated that the (1) This investigation was supported by National Science Foundation Grants GP 1354 and 5978. This is part IX of the series, Organophosphorus Chemistry. (2) To whom correspondence should be addressed a t the Belfer Graduate School of Science, Yeshiva IJniversity. (3) (a) I. J. Borowitz and R. Virkhaus, J. Amer. Chem. Soc., 86,2183 (1963); (b) I. J. Borowitz, K. C. Kirby, Jr., and R. Virkhaus, J . 070. Chem., 81, 4031 (1966). (4) (a) F. Ramirea and S. Dershowitz, ibid., 44, 41 (1957); (b) A. V. Dombrovskii and V.I. Shevchuk, Zh. Obahch. Khim., 88, 1263 (1963). (5) I. J. Borowitz and H. Parnes. J . Org. Chem., 84, 3560 (1967).
(6) H. Parnes, Yeshiva University, unpublished results.
reactions of certain a-halo ketones such as 2-bromodimedone and the a-halobenzyl phenyl ketones (desyl halides) with triphenylphosphine can involve the formation of enol phosphonium salts.7s8 Enol phosphonium salts including 7-10 have been isolated from the reaction of triphenylphosphine with chlorobenzhydryl phenyl ketone 3: the Corresponding bromo ketone 4,'08 dibromobenzyl phenyl ketone 5,'Ob dibromopropiophenone (6), and from other adihalo ketoneslo8 (Scheme I ) . It has been suggested that enol phosphonium salts may arise via displacement by triphenylphosphine on halogen of an a-halo ketone to give an enolate halotri(7) I. J. Borowitz and L. I. Grossman, Tetrahedron Lett., 471 (1962). (8) D. B. Denney and L. C. Smith, J . O w . Chem., 47,3404 (1962). (9) R. D. Partos and A. J. Speziale, J . Amer. Chem. Soc., 87, 5068 (1965).
(IO) (a) K. C. Kirby, Jr.. Yeshiva University, unpublished results; (b) P. E. Rueak, Yeshiva University, unpublished results.
1596 BOROWITZ, RUSEK,AND VIRKHAUS
The Journal of Organic Chemistry
TABLEI THE REACTIONS OF c~-BROMOBENZYL PHENYL KETONEWITH TRIPHENYLPHOSPHINE Time
Reaction conditions
cu-Ketophospbonium bromide
Enol phosphonium bromide
Yields, %"Deoxybenzoin
Diphenylacetylene
Triphenylphosphine oxide
24 hr Benzengsc (0.13M) 58 34-3 4 44 Benzene-methanolb 23 hr 0 lOOd 100 Acetonitrile* 4 days 29 49 19 24 hr Glymeb (1.3M ) 79 15 2 weeks Glyme, room temp 13 73/80 Acetonitriles (0.16M) 7 days 21 70f 9 Nitromethanee 7 days 15 70'80 15 24 hr 100f 100f Glyme-methanole loof 1001 24 hr Ni tromethane-methanola 100' 24 hr 1001 Acetonitrile-me thanol* 1.0-1.1 equiv of PPha was used. Isolated yield unless otherwise indicated. b At reflux. a Same result from reaction in glyme By nmr analysis. 0 Actually isolated. (24hr) or in toluene (24,48hr) a t reflux. By uv or vpc analysis. a At room temperature. 0
f
SCHEME I 0
O H
I/
PhCCHZBr 1
:x 3-6
I1 I
PhC-CCHa
I
Br 2
O H
II I CsHsC4CaHs I
PPh*
X 11, X = Br 14,X = C1 19,X = OSOzCHa
SCHEMEI1 O H
I1
I
CeHsC-cCaHs
I
fPPha X12a,X = Br b, X = C1 C, X = OSOzCHs
X/CaHs
OhPha
\
Rz 7-10
Results and Discussion
The Reactions of a-Bromobenzyl Phenyl Ketone.a-Bromobenzyl phenyl ketone (desyl bromide) 11 reacts with triphenylphosphine in various solvents at room temperature (0.13-0.16 M) to give small yields (5-21%) of the a-ketophosphonium bromide 12 (Table I). High yields (58-85%) of 12 are obtained a t higher temperature ( 8 0 - 1 1 1 O ) or with higher concentrations of the reagents. The best procedure for obtaining a high yield of 12 (7945%) involves reaction of 11 with triphenylphosphine (1.3 M each) in dimethoxyethane (glyme) at reflux temperature (Scheme11). The room-temperature reactions of 11 with triphenyl-
Ohha or
/c=c\
CsHs H 13a,X = Br Ma, X = C1
3,7, X == C1; RI = Rz = Ph 4,8, X == Br; RI = Rz = Ph 5,9,X == R, = Br; RZ = Ph 6, 10,X = R1 = Br; RZ = CHs
phenylphosphonium ion pair. Such an ion pair could then recombine to give the observed enol phosphonium salt.'J1 The actual mechanism of enol phosphonium salt formation has remained unsettled. It is being investigated by us with the aid of the related reactions of appropriate a-halo ketones with optically active methylpropylphenylphosphine.12 We now report our results on the reactions of triphenylphosphine with the a-halobenzyl phenyl ketones and with a-mesyloxybenzylphenyl ketone. Ketophosphonium and enol phosphonium salts as well as derived products are formed in these reactions.
+
/H
\
X-
/c=c\
CaHs CsH6 13b,X = Br 18b,X = C1
phosphine, as followed by nmr spectroscopy, give the enol phosphonium bromide 13 as the main product (70-95'%). The enol phosphonium bromide exhibits a vinyl proton doublet at 7 3.38 with JPH = 1.8 Hz. The presence of one vinyl proton doublet probably indicates that only one isomer of 13 is present. It is anticipated that if both geometric isomers of 13 were present two vinyl proton doublets with different JPH values would be observed. Two doublets are observed for several enol phosphates derived from the reactions of a-halo ketones with triethyl ph0~phite.l~Thus a-chlorobenzyl phenyl ketone 14 reacts with triethyl phosphite to give a 1:2 ratio of 15b:lSa with vinyl doublets at 7 3.31 ( J = 2.5 Ha) and 3.65 ( J = 1.0 He)." Our evidence indicates that the smaller J H ~ value most probably belongs to the trans isomer 15a. An attempt to convert 13 to 15a and/or 15b directly by treatment with triethyl phosphite was unsuccessful SCHEME I11
0
0 I
I
14
15a
15b 13a or 13b
(11) A. J. Speziale and L. J. Taylor, J. 070. Chem., 111, 2450 (1966). (12) (a) 0. Korpiun and K. Mislow, J. A M . Chsm. SOC.,89, 4784 (1967); (b) D. B. Denney and N. G. Adin, Tetrahedron Lett., 2669 (1966); (c) L. Horner and H. Winkler, ibid., 455 (1965).
(13) I. J. Borowitr, M. Anschel. and S. Firstenberg, J . Org. Chem., 89, 1723 (1967). (14) S. Firstenberg, Yeshiva Univereity, unpublished results.
Vol. 34, No. 6, June 1969
REACTIONS OF TRIPHENYLPHOSPHINE 1597
TABLE11 THE REACTIONS OF ~ C H L O R O B E NPHENYL Z Y L KETONE(14) Reaction conditions
a-Ketophosphonium bromide
Time
WITH
Enol phosphonium bromide
TRIPHENYLPHOSPHINE
Yields, %" Diphenylacetylene
Deoxybenzoin
Triphenylphosphine oxide
Benzene, reflux 20 hr 12bSC 17 31 82 Acetonitrile, reflux 10 days 31.5 14 41 e Glyme, reflux 24 hr 12 78 e Methanold 38 days 41 e e Glymed 14 days 100,f 85' e Acetonitriled 7 days 50f 50f Nitromethaned 7 days 53-61 29 e Gly me-methanold 14 days 86 e Acetonitrile-methanold 14 days 48 32 e Nitromethanemethanold 14 days 56 33 e 5 1.0-1.1 equiv of PPha was used. Isolated yield unless otherwise indicated. Same yield from reaction in toluene (24 hr) at reflux. 0 Unreacted 14 was recovered (25%). At room temperature. e Present by tlc. f Yield by nmr. 0 Actually isolated.
(Scheme 111). The displacement of triphenylphosphine from an enol phosphonium salt with the more nucleophilic diethylphenylphosphine has been noted.15 We have successfully used tributylphosphine in such a displacement (see Experimental Section) . Even though displacement by triethyl phosphite should irreversibly lead to an enol phosphate, the poorer nucleophilicity of the phosphite when compared with that of triphenylphosphinels renders such displacement unlikely. While there is yet no firm evidence for the stereochemical assignment for 13 we favor the trans geometry of 13a on the basis of the observed JPHvalue of 1.8 Hz and mechanistic rea,soning as follows. If enol phosphonium salts are formed via the recombination of enolate halotriphenylphosphonium ion pairs then trans geometry is expected. for 13 since the enolate of deoxybenzoin should be more stable in the trans configuration and should react in this configuration (Scheme IV). SCHEME IV
011
+ PPh3 -.
\ C&s/-\
/c&H5
+
BrPPh, -4 13a
H
A similar argument can be made for the conversion of 14 into 18a. While the room temperature reactions of 11 with triphenylphosphine give only 12 and 13 the higher temperature reactions give 12 and several products derived from 13, i.e., deoxybenzoin (16) and diphenylacetylene (17). A number of enol phosphonium salts have been postulated to be or actually are solvolyzed by water to give the corresponding k e t ~ n e . ~ J ~ We J~ believe that the presence of 16 in our reactions is due to the hydrolysis of 13 by residual moisture in the reaction system or by hydrolysis during work-up. The formation of 17 is of special interest in that it indicates a synthetic use for enol phosphonium salts. We7 and o t h e r ~ ' ~have J ~ postulated that the previously o b s e r ~ e d ~formation ~J~ of 17 from 14 is via the enol phosphonium chloride 18. We demonstrate as will (15) H. Hoffmann, Angew. Chem. Intern. Ed. Engl., 8. 737 (1964). (16) G. Aksnes and D. Aksnes, Acto Chem. Scand., l S , 3 8 (17) H. Hoffmann and H. J. Diehr, Tdmhedron Lett., 583 (18) S. Trippett, J . Chem. SOC.,2337 (1962). (19) S. Trippett and D. hL Walker, ibid.. 2976 (1960).
(1964). (1962).
follow that 17 does form from 18, although to a small extent. Our evidence is sufficient to state that the formation of 17 from 11 is via 13 (Scheme V) . SCHEME V
16
13,X~Br 18,X = CI C6r,CdC&
+
OPpba
+
HBr
17
Thedebromination of 11 to 16 with triphenylphosphine in the presence of methanol is quantitative in nonpolar or polar solvents (Table I). The debromination of a-bromo ketones by triphenylphosphine and a protic species has been found to be a general r e . e ~ c t i o n ~ ~ ~ J ~ * ~ ~ failing only in the case of a-bromocamphor which presents an especially hindered s i t u a t i ~ n . ~ ~ , ~ ~ ~ Our most recent work6 indicates that the debromination reaction is acid catalyzed in several cases including 2 and 11 and that it most probably involves a change of mechanism from the S N type ~ of pathway which is involved in a-ketophosphonium salt formation from 1 or (Scheme VI). Z51*
SCHEME VI
+OH OH
I C,H,C=CH,
0
II
+
BrPPha
----t
C6H5CCH3
+
RBr 4- OPPh,
It now appears that the debromination reaction probably involves attack by phosphine on the bromine of a protonated a-bromo ketone.21 The Reactions of a-Chlorobenzyl Phenyl Ketone (14).-The reactions of 14 with triphenylphosphine are summarized in Table 11. Enol phosphonium chloride (20) I. J. Borowitz. K. C. Kirby, Jr., P. E. Rusek, and R. Virkhaus, J . Org. Chem., 88, 3686 (1968); (b) G. Gonis, Lahigh University, unpublished results. (21) Our kinetic studies on the debromination reaction are in progress and will be reported elsewhere.
1598 BOROWITZ, RUSEIC,AND VIRKHAUS 18 is the only product in glyme at room temperature. Reaction in polar solvents, acetonitrile or nitromethane, a t room temperature gives 50-6170 yields of the ketophosphonium chloride 12b. The trend toward higher yields of ketophosphonium salt in reactions in polar solvents or those done at higher temperatures in nonpolar solvents noted for 11is also evident for 14. Thus reaction of 14 with triphenylphosphine in glyme, benzene, or toluene at reflux gives 12% 12b as compared with none in glyme at room temperature. Reaction of 14 with triphenylphosphine in the presence of protic species generally leads to about the same yields of 12b as do the corresponding reactions in the absence of a protic species. It appears that the lack of dehalogenation in this case is related to the reactions of a-chloracetophenone,3a a-chloroacetone,b or a-chloropropiophenonelob wherein the yields of a-ketophosphonium chloride are not significantly decreased by the initial presence of a protic species; i.e., displacement of chloride ion occurs even in the presence of protic species. The yields of desoxybenzoin observed in all of the reactions probably arises from the secondary hydrolysis of the enol phosphonium chloride which forms to some extent under all of the conditions studied. Evidence for the presence of the enol phosphonium chloride in room temperature reactions is based on proton nmr measurements and direct isolation of 18 (see the Experimental Section) . Diphenylacetyleneis a minor product in the reactions of 14 with triphenylphosphine. We have not obtained the high yield (goyc)of 17 from 14 as noted by Trippett and Walker.ls The recovery of starting material in the benzene reaction indicates that 14 reacts more slowly with triphenylphosphine than does 11. This reactivity difference of bromo ketone > chloro ketone has been noted for numerous sets of a-bromo and a-chloro ketones in reaction with triphenylphosphine.6p2z We note that the presence of electron-withdrawing groups on the carbon bearing the halogen of an a-halo ketone enhances enol phosphonium salt formation. Thus, while chloroacetophenone and chloropropiophenone react with triphenylphosphine to give ketophosphonium chlorides, 14 gives both enol and ketophosphonium chlorides and a-chlorobenzhydryl phenyl ketone (3) gives only enol phosphonium chloride? A similar distribution of products exists for the corresponding a-bromo k e t ~ n e s . ~ J ~ ~ We believe that the above results are best explained by mechanistic pathways involving S N type ~ of displacement of halide ion for the ketophosphonium halides, and attack on halogen followed by recombination of the resultant ion pair for the enol phosphonium halides.13 The latter pathway will be elaborated upon in the next section of our Discussion. Attack on halogen is electronically enhanced by electron-withdrawing phenyl groups while displacement of halide ion is sterically retarded by them. The major difference between the a-bromo ketone (22) R. F. Hudson, “Structure and Mechanism in Organophosphorus Chemistry,” Academic Press, Inc., New York, N. Y., 1965, pp 146-151. (23) Our results with optically active methyl%-propylphenylphosphme agree with these conclusions and will be published elsewhere.
The Journal of Organic Chemistry and the a-chloro ketone series is that the initial presence of protic species causes all of the a-bromo ketones to be debrominated completely while the chloro ketones still give a-ketophosphonium chloride formation. We presume that attack on bromine by triphenylphosphine is a more facile process than attack on chlorine since the latter is less polarizable, Le., not so “soft” as is bromine.*‘ The Reactions of a-Mesyloxybenzyl Phenyl Ketone (19).-Reaction of benzoin with methanesulfonyl chloride and triethylamine gives 19 in 69% yield. Reaction of 19 with triphenylphosphine in glyme at reflux gives the a-ketophosphonium mesylate 12c in 81% yield. No other products are noted in this reaction which is the method of choice for the preparation of 12. The yield of 12c is not significantly decreased (80%) when 19 is treated with triphenylphosphine in a 4: 1 mixture of glyme-methanol. Thus these reactions, as those of other a-keto mesylates with triphenylphosphine, involve simple displacement of mesylate ion. The a-keto mesylate reactions thus serve as simplified models for the more complex reactions of the comparable a-halo ketones with triphenylphosphine. The lack of enol phosphonium salt formation from 19, when compared with the behavior of 11 and 14, again suggests that attack on halogen is involved in enol phosphonium salt formation. This is especially so since 19 reacts more slowly with triphenylphosphine than does 11 or 14; Le., the possibility that the clean formation of 12c from 19 merely involves a faster sN2type reaction than is found for 11 or 14 is eliminated. It appears that other pathways are available to 11 or 14 that are not available to 19. The most likely of these pathways is attack on halogen (Scheme IV) . Attack on oxygen by “soft” phosphorus, while found in certain cases where there is really no competitive alternati~e,”~is an unlikely process for monohalo ketones, This is due to the lack of polarization for the oxygen atom which is ((hard” as compared with the polarizable or “soft” bromine or chlorine at0m.2~ Enol phosphonium salts may conceivably be formed by or attack of triphenylphosphine at carbonyl oxygen18~26 by addition to carbonyl carbon and rearrangement of the phosphorus moiety to oxygen (as suggested by us for the formation of certain enol pho~phatesl~).It is difficult to see why these pathways should be completely absent for the mesylate 19 in comparison with the a-halo ketones 11 or 14. Reactions of the Enol Phosphonium Salts.-The addition of water to a mixture of 12 and 13 causes the disappearance of the vinyl proton doublet a t T 3.38 and the appearance of the methylene singlet of deoxybenzoin (16) a t T 5.7 as followed by nmr spectrometry, i.e., the enol phosphonium salt 13 is readily hydrolyzed by water to 16. No reaction occurs if 13 is heated at reflux in glyme for 24 hr or kept at room temperature for 1 week. A similar recovery of 13 occurs if it is treated with triphenylphosphine in glyme at reflux or room tempera(24) R. G. Pearson and J. Songstad, J . A m . Chem. scc., 89. 1827 (1967). (25) For examples of organophosphorus reactions which are likely to involve addition to carbonyl oxygen, see (a) F. Ramirer, Accounts Cham. Res., 1, 168 (1968); @) I. J. Borowitz and M. Anschel, Tetrohedmn Lett., 1517, 5032 (1967). (26) F. Rsmirez, I(. Tssaka, N. B. Desai, and C. P. Smith, J . Ow. Cham., 8% 25 (1968).
Vol. $4, No. 6,June 1969
REACTIOKS OF TRIPHESYLPHOSPHINE 1599
ture for 3 days. There is no formation of diphenylacetylene nor any conversion of 13 into the ketophosphonium salt 12. Diphenylacetylene (17)is formed in 17YGyield from the treatment of 13 with acetonitrile at reflux for 64 hr and in 28% yield from similar treatment of 18. These results indicate that the enol phosphonium salts 13 and 18 can be decomposed to diphenylacetylene as we had originally suggested. The origin of 17 in some of the reactions of 11 and 14 with triphenylphosphine is thus explained. The Role of Ketoylide 20.--Conversion of 12a into a-phenylphenacyltriphenylphosphorane(20) is accomplished in 88% yield upon treatment with sodium methoxide in methanol at room temperature (Scheme VII) . Treatment of 20 with acetonitrile at reflux gives no reaction. It has been previously noted by Trippett and Walkerlg that 20 decomposes to give 17 under pyrolytic conditions at 300". Our control experiment indicates that 20 is not involved in the formation of 17 under normal reaction conditions in solution. Furthermore 20 is reasonably stable to hydrolysis2' and it cannot reasonably be the precursor of desoxybenxoin in the reactions outlined in Tables I and 11. As a further control, ketophosphonium chloride 12b was shown t o be stable to acetonitrile at reflux. No conversion t o any other species was found. Thus the enol and keto phosphonium salts do not interconvert and seem t,o be forming via separate pathways. SCHEME VI1
O H
'hPh, 12a
0-
Pr-
iPh, 20
Experimental Sectionz8 All of the solvents used were distilled from phosphorus pentoxide, calcium hydride, or lithium aluminum hydride. Reactions were conducted under an atmosphere of dry nitrogen. Organic solutions were dried over magnesium sulfate. a-Bromobenzyl phenyl ketone was prepared by the bromination of deoxybenzoin in ether in 84% yield, mp 54.5-56' (from EtOH), lit.20 mp 54-56'. a-Chlorobenzyl phenyl ketone was obtained from Aldrich Chemical Co., mp 66.568" (from hexane), or prepared as in the literature? mp 66.0-67.0", lit.30mp 66-67'. a-Mesyloxybenzyl Phenyl Ketone.-Methanesulfonyl chloride (2.15 g, 0.0189 mol) in dry benzene (40 ml) was added dropwise over 1 hr to a well-stirred mixture of benzoin (4.0g, 0.0189 mol) and triethylamine (3.82 g, 0.0378 mole) in benzene (20 ml) After 1 hr of stirring and removal of triethylamine hydrochloride by filtration, the resultant solution was washed with water, dried, evaporated in vacuo, and recrystallized from EtOAc and from cyclohexane to give 19 as a white solid (3.80 g, 0.0131 mol, 69%), mp 120-121". The ir spectrum (CH2C12) exhibited peaks at 5.89 (C=O), 7.40, and 8 . 5 0 ~(OS02CHs) ;nmr (CDCls), singlets a t T 6.95 (OS02CHa) and 3.15 (COCH) and a multiplet at 2.0-2.9 (10aromatic H).
.
(27) H. J. Bestniann and B. Amason, Chem. Ber., 96, 1513 (1962). (28) The instrumental and other techniques used have been recorded previously.3b (29) H. Limpricht and €I. Schwanert, Ann., 155,59 (1870). (30) A. M. Ward, "Organic Syntheses," Coll. Vol. 11, John Wiley & Sons, Inc., ?'-en- York, N. Y., 1943, p 159.
Anal. Calcd for ClaH1404S: C, 62.06; H, 4.85; S, 11.04. Found: C, 62.18; H,,5.00; S, 10.91. The Reaction of a-Bromobenzyl Phenyl Ketone with Triphenylphosphine. A. In Nonpolar Solvents at Reflux.-a-Bromobenzyl phenyl ketone (2.00 g, 0.00727 mol) and triphenylphosphine (2.02 g, 0.00769 mol) were heated at reflux in benzene (55 ml) for 24 hr. Insoluble white material (2.70 g) was extracted with hot benzene to leave a-phenylphenacyltriphenylphosphonium bromide (12a) (2.26 g, 0.0042 mol, 58%), mp 239.5-241' dec, as an insoluble residue. The ir spectrum (KBr) exhibited peaks at 6.08 ( s ) ,6.97 ( s ) ,7.81 (m), 8.30 (m), 9.18 (s) and 10.06 p (m); nmr (CDCla), multiplets at T 1.5-3.0 (aromatic H ) and a doublet centered a t 1.04 (methine H, Jslpg. = 12.5 Ha). The analytical sample had mp 243-244" dec (from aqueous EtOH). Anal. Calcd for C32H260PBr: C, 71.51; H, 4.88. Found: C, 71.72; H, 5.08. The benzene extract gave crude triphenylphosphine oxide (0.88 g, 0.00317 mol, 44%), mp 149-155", identified by tlc and its ir (CHCb) spectrum on a cyclohexane-insoluble fraction. The cyclohexane-soluble fraction gave diphenylacetylene (0.05 g, 0.00028 mol, 4%), identified via its uv absorption maximum a t 292 mp and desoxybenzoin (a. 0.50 g, 0.0025 mol, 340/,), identified via tlc, ir, and mixture melting point (as below). Similar reactions in glyme (24 hr) or in toluene (24 or 48 hr) gave 12a in 58% yield. Reaction of more concentrated solutions of a-bromobenzyl phenyl ketone and triphenylphosphine (1.3 M each in dry glyme a t reflux for 6 hr) gave 12a in 7945% yield with desoxybenzoin and triphenylphosphine oxide as the other products. B. In Acetonitrile at Reflux.-A mixture of a-bromobenzyl phenyl ketone (2.99 g, 0.0109 mol) and triphenylphosphine (3.09 g, 0.0118 mol) in acetonitrile (70 ml) was kept a t reflux for 4 days. Removal of the solvent in uucuo gave an oil which was slurried in glyme to give 12a (1.70 g, 0.00315 mol, 29%), mp 239.5-241.0" (ir, nmr spectra as above). The residual mixture was dried in uucuo, dissolved in benzene (3 ml), and chromatographed on a column of silica gel (100 9). Elution with benzene (progress monitored via tlc) gave diphenylacetylene (0.362 g, 0.00207 mol, 19%), mp 58-59.5", mmp 57.5-59.5'. Further elution with benzene gave deoxybenzoin (1.05 g, 0.0054 mol, 49%), mp 54-56' (from methanol), mmp 53-56'. C. In Various Solvents Containing Methanol.-A mixture of or-bromobenzyl phenyl ketone (3.00 g, 0.0159 mol), triphenylphosphine (2.85 g, 0.00159 mol), and methanol (1.50 g, 0.0477 mol) in acetonitrile (20 ml) was stirred a t room temperature for 24 hr. After removal of the solvent in vacuo, the nmr spectrum of the residue indicated triphenylphosphine oxide and deoxybenzoin (100% yield); nmr (CDCb), a multiplet a t T 1.7-3.0 (25 aromatic H) and a singlet at 5.80 (2 methylene H). The same results were obtained from reactions in methanolbenzene a t reflux and methanol-glyme or methanol-nitromethane a t room temperature. Work-up of the benzene-methanol reaction by chromatography on acid-washed alumina gave (1) deoxybenzoin (99-100%) via elution with benzene and (2) triphenylphosphine oxide, mp 149-157" (loo%), via elution with 95% ethanol. D. In Aprotic Solvents at Room Temperature.-a-Bromobenzyl phenyl ketone (6.0g, 0.0218 mol) and triphenylphosphine (6.06 g, 0.0231 mol) were stirred in glyme (I5 ml) a t room temperature until the lack of a precipitate with mercuric chloride indicated the absence of triphenylphosphine3b (2 weeks). The resultant solid was filtered under nitrogen and dried in uucuo to give a mixture of the ketophosphonium bromide 12a and the enol phosphonium bromide 13 (10.2 g, 0.0187 mol, 86%). The mixture consisted of 15% 12a and 85% 13 corresponding to a 13% yield of 12a and a 73% yield of 13: ir (CH&12), 3.03.5, 6.00 ( C = O ) , 7.0, 9.0, and 9.6-10.4 p ; nmr (CDCb), a doublet centered a t 7 0.75 (methine H of 12a, J U ~=H12.5 Hz), a multiplet at 1.5-3.0 (25 aromatic H), and a doublet centered at 3.38 (vinyl proton of 13, J 3 1 p ~= 1.8 Hz). Similar results were obtained in acetonitrile and in nitromethane a t room temperature (Table I), The Reaction of a-Chlorobenzyl Phenyl Ketone with Triphenylphosphine. A. In Aprotic Solvents at Reflux.--cr-Chlorobenzyl phenyl ketone (14) (11.30 g, 0.0491 mol) and triphenylphosphine (13.10 g, 0.050 mol) were heated at reflux in dry glyme (15 ml) for 24 hr to give a-phenylphenacyltriphenylphosphonium chloride (12b) (2.9 g, 0.0059 mol, 12%): mp
1600 BOROWITZ, RXSEK, AXD VIRKHAUS 237-240"; ir (CH&12), 6.0 (C=O), 6.95, and 9.06 p ; nmr (CDCL), a multiplet at T 1.4-3.1 (25 aromatic H ) and a doublet centered a t 0.35 (methine H, J ~ ~ = PH 12.5 Hz). Thin layer chromatography showed triphenylphosphine oxide and deoxybenzoin (isolated in 78% yield) to be the only other species formed. The same yield of 12b was obtained from reactions in benzene or in toluene (Table 11). Similar reaction of 14 (1.32 g, 0.0057 mol) in acetonitrile (40 ml) a t reflux for 10 days, followed by a work-up as for the reaction of 11 (above), gave 12b (0.90 g, 0.0018 mol, 31.5%), mp 237-240", diphenylacetylene (0.143 g, 0.00081 mol, 14%), mp 57-59' [from petroleum ether (bp 30-60")], deoxybenzoin (0.462 g, 0.0023 mol, 41%), and triphenylphosphine oxide. B. In Aprotic Solvents at Room Temperature.-Reaction of 14 (11.3 g, 0.0491 mol) and triphenylphosphine (13.0 g, 0.050 mol) in glyme (15 ml) for 14 days a t room temperature (reaction complete by lack of precipitate with mercuric chloride) gave the enol phosphonium chloride 18 as a white solid which could be filtered under nitrogen from the mixture (20.9 g, 0.0425 mol, 85% isolated yield, 100% yield by nmr) : ir (CH2C12),3.2-3.5, 7.0, 9.0, 9.9, and 10.1 p ; nmr (CDCla), a multiplet a t T 1.7-3.0 (25 aromatic H ) and a doublet centered a t 3.45 (1 vinyl H, J 3 1 p ~= 2.0 Hz). Similar reaction of 14 and triphenylphosphine (0.050 mol each) in acetonitrile (15 ml) (Table 11) gave, after removal of the solvent in uucuo, a 1 :1 mixture of enol and ketophosphonium salts as an oil: nmr (CDCls), a multiplet a t T 1.7-3.0 (25 aromatic H) and doublets centered a t 3.45 (vinyl H of 18, Jaip~= 2.0 Hz) and 0.35 (methylene H of lZb, J a i p ~= 12.5 Hz). Similar reaction in nitromethane gave 14 (53-61%) and deoxybenzoin (as%),mp 54-56'. No diphenylacetylene was found in these reactions (tlc). C. In Methanol or Methanol-Containing Solvents.-Reaction of 14 and triphenylphosphine (0.017 mol each) in methanol (20 ml) for 5.5 weeks a t room temperature gave, after removal of the solvent in uuouo, an oil which was slurried in glyme to give 12b (3.44 g, 0.007 mol, 41%), mp 238.5-240.5". The residue contained triphenylphosphine oxide and deoxybenzoin (by tlc on silica gel using 5% EtOAc-CsHg for development). Similar reaction and work-up as for 11 for reactions done in methanol-contairiing solvents gave yields indicated in Table 11. The Reactions of a-Mesyloxybenzyl Phenyl Ketone with Tripheny1phosphine.-Reaction of 19 and triphenylphosphine (0.00655 mol each) in glyme (15 ml) a t reflux for 24 hr gave 12c 8s a white powder (3.05 g, 0.0053 mol, 81%): mp 247248.5'; ir (CHZCI,), 5.98 ( C 4 ) and 8.3-8.5 p (-0S02CHs); HI nmr (CDCL), a multiplet a t T 1.6-3.1 (25 aromatic H and 1 methine H ) and a singlet a t 7.25 (3 H of OS02CHa) ; nmr (CDCla), -25.6 ppm (HaPO4 = O).31 A similar reaction in 4: 1 (v/v) glyme-methanol gave 12c in 80% yield. The Conversion of the Enol Phosphonium Halides 13 and 18 into Diphenylacety1ene.--A 1 :1 mixture of the ketophosphonium chloride 12b and the enol phosphonium chloride 18 (a total of 0.022 mol) was heated at reflux for 64 hr in acetonitrile (500 ml) to give recovered 12b (4.87 g, 0.0099 mol, 45.5%), mp 239-241', (31) The alp nmr spectrum was kindly determined by Jeolco, Inc., a t 24.29 and at 40 SfHz on JNM C-GOHL and C-100 spectrometers.
The Journal of Organic Chemistry mmp 238-241.5'. The residual oil was chromatographed over silica gel (200 g) Elution with benzene gave diphenylacetylene (0.54 g, 0.0030 mol, 28% based on 18), mp 57-57.5", mmp 57.540". The reaction mixture also contained deoxybenzoin and triphenylphosphine oxide (tlc) Similar treatment of the enol phosphonium bromide 13 gave diphenylacetylene (17%), deoxybenzoin, and triphenylphosphine oxide. Reaction of 13 in glyme a t reflux with or without the presence of triphenylphosphine (1 equiv) for 3-7 days led to recovered 13 in 82 and 95% yields, respectively. The Hydrolysis of the Enol Phosphonium Bromide 13.Water (4 drops) was added to a solution of 13 (0.602 g, 0.0011 mol) in CDCh (1 ml) in a nmr tube. After 5 min of shaking, 13 was completely hydrolyzed since the nmr spectrum showed the absence of the vinyl proton of 13 at T 3.3 and the presence of the methylene protons of desoxybenzoin a t 5.8. Attempted Reaction of 13 with Triethyl Phosphite.-The enol phosphonium bromide 13 (2.0 g, 0.0037 mol) and triethyl phosphite (0.616 g, 0.0037 mol) were heated a t reflux for 3 days in dry benzene (10 mi, distilled from LiAlH,) to give a black mass which contained triethyl phosphate, triphenylphosphine oxide, and deoxybenzoin (tlc). The Reaction of 13 with Tributy1phosphine.-Tributylphosphine (0.189 g, 0.000934 mol) was added to 13 (0.501 g, 0.000934 mol) in CDCla (1 mi). Triphenylphosphine was formed immediately (by tlc and formation of the adduct with mercuric chloride). The Formation of a-Phenylphenacyltriphenylphosphorane (20) .-Sodium methoxide (0.21 g, 0.0039 mol) in methanol (10 ml) was added to a stirred solution of 12a (2.00 g, 0.0037 mol) in methanol (25 ml) and stirring was continued for 30 min.'b The resultant solution was poured into water (200 ml) to give a white precipitate which was washed with water and recrystallized twice from EtOAc to give 20 (1.5 g, 0.0033 mole, 88%) : mp 19.5197" (lit.28mp 192-194') ;uv, A z o " 320 (e 7200), 275, 267.5, 262 mp; ir (CHCL), 6.68 (vs), 6.75 (vs), 6.95 (s), 7.23, 8.85, 9.04 (s), 9.33, 9.73, 10.0, and 10.33 p ; tlc (25% MeOH-CeH6), Rt 0.74 us. 0.59 for triphenylphosphine oxide. A solution of 20 (0.253 g, 0.00055 mol), mp 190-195', in acetonitrile (10 ml) was heated a t reflux for 24 hr to leave only 20 (0.25 g, loo%), mp 187-194'. The Stability of a-PhenylphenacyltriphenylphosphoniumChloride in Acetonitrile at Reflux.-A solution of 12b (0.30 g, 0.00061 mol) in acetonitrile (10 ml) was heated at reflux for 24 hr to give recovered 12b (0.31 g, loo%), mp 233-238", ir (CHCla) identical with that of genuine 12b.
.
.
Registry No.-Triphenylphosphine, 603-35-0; 1484-50-0; 12a, 1530-47-8; 12b, 19254-98-9; 19254-99-0; 14, 447-31-4; 19,19255-01-7.
11,
12c,
Acknowledgment.-We are indebted to Professor Claibourne E. Griffin, University of Pittsburgh, for the early nmr spectra and to Professor Grace Borowitz, Upsala College, for stimulating discussions and aid in obtaining 31Pnmr spectra.
